Sept., 1938
Tim STANDARD GLECTRODE POTENTIAL OF SILVER
ride was formed during the course of their electrometric titrations, the electrodes were coated with the electrolytic variety. If the concentration of silver chloride in the immediate neighborhood of the electrodes was successfully controlled by the electrolytic variety, the results should be comparable with those of the present research. Interpolation based on Fig. 3 leads to 1.577 X in excellent agreement with the above, but the value for the solubility in pure water reported by Brown and MacInnes is 1.314 X which is 1.5% lower than that given in Table 111. The discrepancy is caused by the difference between the activity coefficientsused by Brown and MacInnes13 and those which may be obtained from consideration of equation (6) and Fig. 3. The slope of the 25' plot in Fig. 3 is 0.051, which leads to the expression l o g y = -0.506 1/; 0.43~ (10) for the activity coefficient of silver chloride. On the basis of this equation, the result of Brown and MacInnes would be increased to 1.332 X Although this concordance leaves little to be
+
(13) Values obtained by Neuman [THISJOURNAL, 64,2196 (1932)l from solubility measurements in which presence of precipitated silver chloride was determined by Tyndall beam.
2233
desired, it should be borne in mind that two electrometric determinations of the solubility of identical samples would not be strictly comparable unless the extra-thermodynamic hypotheses used to eliminate liquid junction potentials could be verified experimentally, or shown to be equivalent. It therefore seems necessary to reserve final judgment upon the absolute accuracy of the results reported in this paper until the reliability of the hypothesis represented by Fig. 2 has been established, or discredited, by further experimental work. Some familiar indirect checks upon the accuracy of solubilities and their temperature coefficients will be considered in a later communication.2
Summary A method is proposed by which heterionic liquid junction potentials may be eliminated by extrapolation. Experimental data are presented which indicate that the junction extrapolation is linear as a first approximation. The method is applied to the determination of the solubility product of electrolytic silver chloride in water a t 5 , 15, 25, 35, and 45'. NEWHAVEN, CONN.
RECEIVED JUNE 11, 1938
lCONTRIElUTION FROM THE STERLING CHEMISTRY IiABORATORY OF Y A L E UNIVERSITY ]
The Elimination of Liquid Junction Potentials. 11. The Standard Electrode Potential of Silver from 5 to 4S0, and Related Thermodynamic Quantities BY BENTONBROOKSOWENAND STUART R. BRINKLEY, JR.~ In the first paper of this series,2an extrapolation method for eliminating heterionic liquid junction potentials was described, and employed in a determination of the solubility product of silver chloride. It was considered desirable to test the validity and generality of the method by applying it to a comparison of the silver and hydrogen electrodes in buffer solutions. The results of these two investigations are interrelated through the potential of the silver-silver chloride electrode, and can be used to check one another. By the use of identical measuring equipment and technique, and consistent supplementary data, the con-' cordance of these results becomes a necessary, but (1) This communication embodies part of the experimental material to be presented by Stuart R. Brinkley, Jr., to the Graduate School of Yale Universityin partial fulfilment of the requirements for the degree of Doctor of Philosophy. (2) Owen, THISJOURNAL, 60, 2239 (1938).
not a sufficient, condition to the validity of the extra-thermodynamic extrapolation upon which the method is based. Complete demonstration of its validity is contingent upon further direct investigation of the extrapolation, and comparison of its consequences with thermodynamic data obtained without recourse to liquid junctions. Several such comparisons can be made with the data at hand, and others will be available in the future. Buffer solutions were used to depress and stabilize the hydrogen ion concentration, so that transfer of electricity across the liquid junction would be accomplished, for the most part, by ions other than hydrogen. The cell may be represented as follows HZI HAC(,), NaAct,) I HAC(,), NaAc(, -mz). AgAc(,,) I Ag+ The ionic strength, p v m, is practically identical
BENTONBROOKS OWENAND STUART R. BRJNKLEY, JR.
2234
on both sides of the junction, but a small fraction, x , of the total salt in the right-hand solution is
silver acetate. The electromotive force of this cell is given by E = -Eo Ary - k log a d &
* Ej
(1)
where k = 0.00019S.WT, and aH and a i g denote the activities of the hydrogen and silver ion in the left and right-hand solutions, respectively. The ionization of acetic acid requires that (2)
aH = KaixAo/aAc
and by substitution, equation (1)becomes E = -E,&
+ k log a.&A,!aHA,
- k log K
* E,
(3)
With a buffer ratio unity, it is permissible to write xEX= K throughout the concentration range investigated. Accordingly equation (3) becomes E - k log M X ( V I + K ) / ( w z- K) + k- log k =
+ k log
YAgYAe/YHAc
* E, (4)
by rearrangement, and introduction of concentrations and the corresponding ionic activity coefficients. It is evident that when x = 0 the solutions are identical, and Ej vanishes. At constant m, a plot of the left-hand member against some convenient function of x yields the intercept [ E - k log W Z X ( V Z .K)/(nt - K) 4- k log K ] z q
+
-E&
+ k log
Y A ~ Y A J Y H A ~ (5)
Vol. 60
copper turnings at 700". Sodium carbonate was prepared fresh for each cell solution by heating the bicarbonate, in a platinum dish, in an electric furnace a t 290 * 10' for at least ten hours. The calculated amounts of sodium carbonate, silver acetate, acetic acid stock solution and water were weighed into the solution flasks to obtain solutions with the desired valuas of x, m and a buffer ratio of unity. Conducrivity water was used, and vacuum corrections applied. The solutions for the hydrogen halfcell were swept with hydrogen which had passed through a saturator of large volume, and they were then kept under hydrogen. The silver half-cell solutions were swept with nitrogen, and stored under this gas. The silver solutions were protected from light. The essential features of the cell have been described by Owen.2 X o intermediate solution was required, and our cell differs from his only in the elimination of one of the junction stopcocks, and the necessary modifications for the electrodes employed. The silver electrodes were prepared by the thermal decomposition of silver oxide a t 480". The sample of silver oxide which was used was the same as for the previous investigation.2 The hydrogen electrodes were of the usual type employed in work from this Laboratory. The measuring equipment was the same as that of the previous investigation, and the technique was practically identical. The cells were filled twelve hours before the fist measurement, the usual vacuum technique being employed. Measurements were made from 5 to 45' a t 10' intervals, and the cells were refilled frequently during the course of the measurements.
Discussion of the Results
Since the two solutions axe identical, it is no longer necessary to retain the prime on the activity coefficient term. At high dilutions, this term becomes equal to -2ka +, where a is the DebyeHiickel limiting slope. Therefore we may add 2 k a d/i7i to the values of the bracketed term of equation (5), and extrapolate3the resulting quantities to zero ionic strength. The intercept is
The observed electromotive forces of the cells were corrected to a partial pressure of hydrogen of one atmosphere, and to round temperatures. They are recorded in Table I. The reported values are the mean of two cells (of four cells, in a few cases) differing by no more than 0.03 mv. In one case (vz = 0.05, 3c = 0.089), the value represents two independent runs. The first group of -Eig. cells were taken from 5 to 4.5' in one day. The Materials and Technique second were taken from 5 to 25', maintained a t Baker c. P. glacialacetic acid was purified further by one partial freezing, and two distillations from chromic oxide 25' for three days, returned to 5' and then taken All values for the two runs agree within in a n all-glass still. The acid was diluted with conduc- to 4.5'. tivity water to form a stock solution approximately 0.2 the deviation noted above. The cells therefore moIa1. It was standardized by comparison with Bureau of appear to be very stable once equilibrium is atStandards potassium acid phthdate and an independently tained. Reproducible results were not obtained standardized hydrochloric acid solution. The titer wa5 for silver concentrations below 0.002 molal, and known to 10.04%. The solution was stored under these results have been discarded. hydrogen. Eimer and Amend c. P. silver acetate was analyzed gravimetrically, as silver bromide, and found to The values for the dissociation constant of acebe 99.45 =t 0.03% silver acetate. It was used on this tic acid were taken from a paper by Harned and basis without further purification. Baker and Adamson Ehlers.4 The left-hand member of equation (4) .dium bicarbonate was recrystallized once from distilled is plotted against x in Fig. 1 at the five temperawater saturated with carbon dioxide, and dried in an atmosphere of carbon dioxide. Commercial nitrogen and tures investigated. The results at 25' are s h i hydrogen were freed from oxygen by passing them over larly plotted on a larger scale in Fig. 2 to illustrate :3) Hitchcock, THISJOURNAI., SO, 2076 (1928)
( 4 ) Harned and Ehiers, a i d . , 65,662 (1933),Equation 7.
THESTANDARD ELECTRODE POTENTIAL OF SILVER
Sept., 1938
TABLE I OBSERVED ELECTROMOTIVE FORCES AND INTERCEPTS FROM THE x
to..
FIRSTEXTRAPOLATION (FIG.1)
... .5'
0.96085 .95428 ,94917 .94101 .93389 ,92612 .80910
0.2299 .1660 .1299 .0892 .0652 ,0475 Intercept
15' 25' 35' 45' m 0.05 0.95477 0.94944 0.94452 0.93977 .94793 .94258 ,93748 .93252 .94265 .93717 .93166 .92659 .93413 .92837 .92283 .91737 .92626 .92083 .91493 .go938 ,91944 .91296 .go696 .go109 .79830 .78805 .77740 .76630 m = 0.03
0.96512 0.95960 0.95461 0.95021 0.94598 .95893 .95323 .Q4807 .94347 .93892 .95312 .94718 .94177 .93675 .93202 ,93990 .93350 .92789 .92234 .91716 .93385 ,92725 .92141 ,91572 .91029 .81070 .80015 .79015 .77950 .76870
0.40 .30 .2299 .1299 .10 Intercept
the accuracy of the extrapolation. In Fig. 2, the radius of the experimental points represents an uncertainty of 0.1 mv. With the exception of a single point a t 5' (m = 0.05, x = 0.047)) the deviations from linearity are in no cases greater than 0.10 mv. This result for cells with buffer solutions corroborates the linearity previously observed2 without buffer solutions. In that investigation, the slope of the extrapolation function increases with increase in ionic strength; in this investigation, the slope has the opposite sign, and shows an algebraic decrease with increasing ionic
-
m 0.02 0.96302 0.95748 0.95248 0.94801 0.94367 .95807 .95235 .94708 .94245 .93796 .95143 .94549 .94006 .93513 ,93042 ,94222 .93583 .93018 .92491 .91983 .93544 .92887 .92295 .91750 .91222 .81195 .80155 .79145 .78105 .77040
0.50 .40 .30 .20 .15
-
Intercept
0.60 .40 .30 .20
m 0.01 0.95285 0.94691 0.94165 0.93657 0.93202 ,94370 .93739 .93175 .92642 ,92146 .93701 .93055 .92468 .91918 .91402 .92748 .92073 .91458 .go878 .go334 .81370 .80335 .79330 .78315 .77255
o.*lwj
Intercept
0
0.2
0.4
0.6
X.
5"
Li
2235
Fig. 2.-Extrapolation to eliminate liquid junction potentials at 25'. Radius of circles is 0.1 mv.: 0, m = 0.05; 0 , m = 0.03; @, m = 0.02; 0 , m = 0.01.
15'
25 O
35O
45 O
1
0.76 0
0.4
0.2
0.6
X.
Fig. 1.-Extrapolation to eliminate liquid junction potentials: 0,m = 0.05; 0 , m = 0.03; a,m = 0.02; 0, m = 0.01.
strength that is more rapid than in the former case. By reason of its steeper slope, the extrapolation a t m = 0.05 is probably of no greater accuracy than that at m = 0.02, although in the former case it was possible to extend the measurements to a lower value of x. It should be noted that a plot of the electromotive forces of these cells against x does not form a basis for the estimation of the magnitude of the liquid junction potential, or of its variation with total ionic strength or temperature, since these plots also contain the variation of the unknown activity coefficient term, log Y A g + Y ~ c - / ~ ~ ~The c. linearity of the extrapolation function seems established within the experimental error for the concentration range which has been investigated with both types of cells.
The intercepts from the extrapolation in Fig. 1 are recorded in Table I to the nearest 0.05 mv. The term 2ka &i is added to these intercepts as explained above, and the resulting quantities ex-
BENTONBROOES OWENAND STUART R. BBINZLEY, JR.
2236
+ 0.801
Vol. 60
We have employed H a r n d and Ehlers’ valuess of because their investigation led to the values of the dissociation constant of acetic acid4 which we have used above. The result of this calculation is given in column three of Table 11. The two sets of data constitute a satisfactory check on the precision of the measurements. The difference between the two sets of values is well within the combined experimental uncertainties of the two determinations. Accordingly, we consider the best values of E l , to be the mean of the two determinations. This can be expressed satisfactorily by the equation
25
I
Eig =
v
‘ . ,
-0.7991
+ 0.000988(t - 25) + 7 X l U w 7 ( t - 25)?
h
( 7) 35O
8 0.789 M
I
I 0 04
li 02
/I
m. Fig 3. -Extrapolation to eliminate activity coefG&nts.
trapolated to mfinite dilution in Fig. 3. The values of E l g thus obtained are recorded in Table TI. Uncertainties in concentrations, temperature TABLE I1 STANDARD ELECTRODE POTENTIAL OF SILVER, AND DERIVED THERMODYNABI’IC QUANTITIES FOR THE ELECTRODE REACTION 1,
oc. 5
15 25 3.5 4R
-Eig
-E:&
-Ea,
Obsd. Ref. 2 Eq. (7) ASo AHo 0.81855 0.8167 0.81868 .8087 ,8690 .SO881 2L46 25,127 ,7990 .?Be2 .79910 22.79 25,221 ,7891 ,7892 ,78915 23.10 26,319 ,7788 ,77915 77906
. .. .
... .
ACp‘
.. 9.30 9.62 9.96
and the measurement of the electromotive force
Values calculated from this equation are given in T a b k 11, column 4. The maximum deviation of observed from calculated values is 0.26 mv., and the average deviation is 0.11 mv. for both series of results. Since the temperature coefficient of E;, is of particular interest, it should be remarked that by a change in the constant term this equation will reproduce either set of values with a maximum deviation of 0.1 mv., and a11 average deviation of 0.03 mv. The values from the equation are used in all subsequent calculations. Lewisb and Noyes and Brann7 have compared the silver electrode in a 0.1 normal silver nitrate solution with the 0.1 normal calomel electrode. On the basis of various values for the reference electrode potential, and of various assumptions regarding the liquid junction potential and ionic activities, different authors have obtained the values -0.799P and -0.7978’ for E i g at 25” from these data. Chloupek and DaneSio h a w compared the silver electrode in a silver sulfate solution with the decinormal calomel electrode, using various bridge solutions. They obtain -0.7996 for E&, at 2.5’. Considering the diversity of the methods used to eliminate liquid junction potentials, the agreement of these values with ours seems satisfactory. Differentiation of equation (7) yields dEi,/dT = 0.000988
+ 1.4 X 10-6(t - 25)
(8)
( 5 ) Harned and Ehlers, THISJOUHNAT 66, 2179 (1933), Equation
of the solubility
d siEver chloride have
relation E.& = k log K
+ E;,n
(6)
(6) Lewis, , b i d , 98, 158 (1906). (7) Noyes and Brann, i b r d , 84, 1016 (1912) (8) Lewis and Randall, “Thermodynamics and the Free Energy of Cbemical Substances,” McGraw-Hill Book Co , New York,N Y , 1923, p. 414; Gerke, Chcm. R e s , 1, 377 (1925) (9) “Internatiooal Critical Tables,” Vol. VI, p 333 (10) Chloupek and Danes, Coil Czech Chrm Comm , 4, 124 (1932)
THESTANDARD ELECTRODE POTENTIAL OF SILVER
Sept., 1938
Skotnickfll has measured the temperature coefficient of various half-cells against reference electrodes maintained a t constant temperature. His method involves a thermal e. m. f. in the bridge solution, and a correction for the change in activity of the electrolyte with temperature, which is estimated from conductance measurements. He obtained the value 0.978 mv./deg. for dEi,/ d T at 25'. Lingane and LarsonI2 have measured the temperature coefficient of the cell Quinhydrone
1
HC104cl I HClOic1, AgNOsg
I Ag
They obtain the value 0.967 mv./deg. for dEig/ d T at 25'. Their lowest concentration, is 0.04 molar, and they made no extrapolation to infinite dilution. Their silver ion concentrations are lower than those with which we could obtain reproducible results, and no account was taken of liquid junction potentials. In view of the nature of the calculations employed by Skotnickf, and by Lingane and Larson. we regard our value, 0.000988, as the most reliable figure.
2237
- logKkBr 5702.2/T - 11.1822 + 0.014610T (10) - log KA,I = 7084,0/T - 12.0389 + 0.014610T Values of the logarithms of the solubility products and the corresponding solubilities are given in Table 111. The solubilities have been converted TABLE111 SOLUBILITY PRODUCTS AND SOLUBILITIES OF THE SILVER HALIDES t, o c .
5 15 25 35 45
-log?:
10.5945 10.1520 9.7492 9.3809 9.0425
x 106 0.506 0.842 1.337 2.038 3.002
A Br -log x 10' -log KAgIc x 10' 13.3860 0.203 17.4980 0.178 12.8201 .380 16.7598 ,417 12.3026 .704 16.0813 .908 11.8280 1.213 15.4562 1.859 11.3920 1.997 14.8792 3.599
2,
into moles per liter of solution in order to facilitate comparison with the literature. Activity coefficients were calculated by the Debye-Hiickel limiting law. The discordance between the values in the literature for the solubility of silver chloride has been discussed in the previous paper.2 The same considerations apply to the bromide and iodide. Comparison with previously recorded results does Derived Thermodynamic Quantities not, therefore, constitute a criterion of the absoBy application of familiar thermodynamic relalute accuracy of the results in Table 111. Our tions to equations (7) and (S), AS", AHa and AC; values do constitute a self-consistent set of soluare computed readily for the electrode reaction bilities which have been derived from e. m. f. Ag H + + '/*Ha Ag+ methods alone. It is important to note in this The values of these quantities are given in Table connection that identical values of the standard I1 for 15, 25 and 3 5 O , the three temperatures at potential of the silver-silver chloride electrode6 which equation (7) best reproduces the experimen- were used in all of the supplementary data13pl4intal results. volved in the construction of Table 111. The soluThe solubility products of the silver halides bilities in Table I11 are 0.25% higher than those may be computed from the relation recorded by the first paper2due to the combination k log K A ~ = X E L - ELx (64 of those data with ours in equation (7). The By substitution of equation (7) from this paper, values in the literature for the solubility of silver and the equations for the temperature variation bromide and silver iodide consist of determinaof the standard silver halide electrode poten- tions a t a single temperature. The mention of t i a l ~ ~ into * ~equation ~ , ~ ~(sa), the following expres- a few of these will be sufficient. The determinasions for the logarithm of the solubility product tion of the solubility of silver bromide a t 21.1 O by are obtained Kohlrausch and Dolezalek16is 5% lower than our interpolated value. That of Bodlander and k log K A ~ C=I -0.57671 0.00034248(t - 25) 2.584 X 10-6(t - 25)' 9.948 X 10-@(t- 25)a Fittigla a t 25' is 2% higher than ours. The disk log K A ~ =B -0.72776 ~ 0.000490(t - 25) cordance with other values is considerably 2.9 X 10-6(t - 25)' (9) greater. The disagreement with the few rek log K A ~=I -0.95129 O.O0066O(t - 25) corded values of the solubility of silver iodide is 2.9 X 10-6(t - 25)* even greater. The value of Goodwin" a t 25' is or 5% higher than ours, and that of ThielI8 a t the - log Kkrj1 5905.8/T - 22.8538 0.0578541"5.0131 X 10-6T' same temperature is 15% higher than ours.
+
+
+ + + +
+
(11) Skotnickf, CoU. Cscck. Chem. Comm., 8, 498 (1936). (12) Lingane and Larson, THISJOURNAL, 69, 2271 (1937). (13) Owen and Foering, ibid., 68, 1575 (1936). Equation 2. (14) Owen, ibid., 67, 1526 (1935). Equation 6.
(15) Kohlrausch and Dolezalek, Bcrl. Sitebey., 1021 (1901). (16) Bodlander and Fittig, 2.physik. Ckcm., 39, 605 (1902). (17) Goodwin, ibid., 13, 646 (1894). (18) Thiel, Z. anow. Chcm., 24, 57 (1900).
2238
RENTON
BROOKSOWEN
AND STUART
R. BRINKLEY, JR.
TABLE IV DERIVED THERMODYNAMIC QUANTITIESFOR AgX +Ag+ X-
+
AgCl
TlIIE
Vol. 60
REACTION
AgBr
AgI
6, 'C.
AS*
AH0
Ace,
4.5 0
Affo
Act
A90
AH0
Ac"P
15
9.19 7.90 6.77
16,022 15,653 15,313
-34.7 -35.5 -36.3
12.64 11.30 9.96
20,542 20,150 19,749
-38.5 -39.9 -41.2
16.56 15.22 13.88
26,865 26,473 26,068
-38.5 -39.9 -41.2
25 35
TABLE V THERMODYNAMIC DATAz3
SUPPLEMENTARY 1,
'C.
Hz 3.43 3.43 3.44 6.62 0 81
I/z
15 25 35
c;
AgCl
Ag
6.09 6.10 6 12 5.65 1 50
12 05 12.14 12.23 9 37 9.29
SO
AgI
AgBr
12.38 12.52 12.66 8.81 14.1
l/z
Hz
15.50 12.87 13.01 15.61 15.73 13.15 8.81 = u 14.1 = h X lo3
TABLEVI RELATIVE IONICENTROPIES AND HEAT 1,
oc.
sb, 16.96 17.38 17.87
15 25
35
S&
-
14.82 13.52 12.30
$31
-
20.84 19.52 18.12
AgI
22.59 23.00 23.40
25.16 25.60 26.03
26.66 27.10 27.53
CAPACITIES
-
c;c,
26.26 24.94 23.54
12.0 12.3 12 6
-34 6 -35.7 -36.7
(19) Pitzer and Smith, THISJOURNAL, 59, 2633 (1937) (SO) Lange and Fuoss, 2.fihrsik. (21) Latimer, Schutz and Hicks, (22) Lange and Shibata, Z.physs (231 The entropy and heat capacity data at 25', except for hydrogen. are f review of Kelley, Bur Mincs, Bull , 394, 3-5 (1935) The heat urphey, J Ckrnr Phrs , 6 , 6 3 i of all heat capadties are (1934), being described by alculated from "b" and c", at 298.1'K. (given in BuLi 394) The entropy at temperatures other than 298.1"K. is then given by S'J = Slpa o(2.303 log I'/ 298 1) b(T 298 1 )
+
AgBr
10.00 10.20 10.40
C L +
+
-
AgCl
S! -
By application of the usual thermodynamic relations to equations (9) or (lo), it is possible to compute ASo, AHo and AC: for the reaction, AgX -+ Ag+ X-. The values of these quantities for the silver halides are given in Table I V for 15, 25 and 35'. Pitzer and Smith1$have calculated the heat of solution a t 25' of silver chloride from the heat of precipitation data of Lange and F ~ o s s . They ~ ~ obtain the value AHo = 13,740 cal. Latimer, Schutz and Hicksz1report 26,710 cal. for the heat of solution of silver iodide, at 25' using the data of Lange and Shibata." The supplementary thermodynamic data riecessaxy for the computation of relative ionic entropies and heat capacities are summarized in Table V.23 Using the convention that the entropy of the hydrogen ion is zero at all temperatures, the relative ianic entropies and heat capacities of the silver and halide ions have been calculated from the values in Tables 11, IV and V. These quantities are summarized in Table V I at 15, 25 and 35'.
+
Ag
C;Br
-
-38.1 -39.6 -41.2
CkI
-
-37.6 -39.2 -40.7
Skotnickf's1l value of dEig/dT leads to 17.0 cal./deg. mole for $?,Ag+ at 25'. Lingme and Larson12 obtain 16.7 cal./deg. mole for the same quantity. A more reliable value from calorimetric methods is from the recent heat capacity data for silver oxide of Pitzer and Smith.lg They at obtained 17.46 * 0.2 cal./deg. mole for gg+ 25'. They have also recalculated the value given by Latimer, Schutz and Hicks,21based upon the heat of precipitation data of Lange and Fuoss,"O and obtained 17.62 A= 0.2. They proposed the mean of these two figures as the best value, but in view of the approximate heats of dilution used and the to correct the data of Lange and FUOSS, close agreement with our value, we suggest the adoption of 17.4 * 0.2 cal./deg. mole as the most probable value of soAg+a t 25'. From a number of sources, Latimer, Schutz and Hicksz1have selected 13.5 * 0.1, 19.4 * 0.4 and 23.7 * 0.7 cal./deg. mole for the best valuesZ4of S $ - , s",,- and Si-,respectively, a t 25'. The agreement with the values in Table VI is well within the estimated uncertainties for the chloride aiid bromide ions, and is probably also satisfactory in the case of the iodide ion. Summary A method for the elimination of heterionic liq(24) While this paper was in press, Latimer, Pitzer and Smith [THIS JOWENAL, 80, 1829 (1938)] have published another review of the subject. Their revised "best" values are 13 5 * 0.1,19.7 f 0.2 and 25 3 * 0.5 for the entropies of the halide ions, and 17 64 * 0.15
[or the silver ion
Sept., 1938
ASSOCIATION IN THE CARBOXYLIC ACIDSBY INFRARED ABSORPTION
uid junction potentials by extrapolation has been tested in cells with buffer solutions. Experimental data are presented which indicate that the junction extrapolation is linear as a first approximation. The method is applied to the determination of the standard potential of the silver electrode a t 5, 15, 25, 35 and 45’. Values of AS”, AHQand AC; have been calculated for the electrode reaction a t 15, 25 and 35’. By combination of the normal potentials of the
[CONTRIBUTION FROM THE CHEMICAL
2239
silver and silver-silver halide electrodes, the solubility products and solubilities of (electrolytic) silver chloride, bromide and iodide have been Values of ASO, AHo computed from 5 to 4.5’. and AC; for the reaction AgX = Ag+ Xhave been recorded a t 15, 25 and 35’. A self-consistent table of relative ionic entropies and heat capacities from electromotive force data has been made for the silver ion, chloride ion, bromide ion and iodide ion a t 15, 25 and 35’. NEWHAVEN, CONN. RECEIVED JUNE 29, 1938
+
LABORATORY OF THE
UNIVERSITY OF ILLINOIS]
Infrared Absorption Studies. V. Association in the Carboxylic Acids BY A. M. BUSWELL, W. H. RODEBUSH AND M. F. ROY The electron conception of valence proposed study of simple solubility by Zellhoefer, Copley some twenty years ago led to theeprediction in and Marvel4 has yielded a surprising amount of 1920 of two types of associated liquids.’ In one additional confirmatory data along these lines. The existence of these two types of association of these the molecules should be combined with the formation of definite polymers (dimers in the has perhaps contributed to the confusion which case of acetic acid) while in the other type poly- for many years characterized the attempts of mers of indefinite molecular weight and structure physical chemists to discuss the behavior of the were anticipated. Radically different behavior, so-called ‘inormal” and “associated” liquids. especially in the case of solvent properties, was pre- This confusion was all the more inevitable because dicted in the two cases. The liquids composed of it is now obvious that the normal or unassociated polymers of indefinite molecular weight comprised liquid lies intermediate in its properties between the typically polar solvents such as water or alco- the two types of associated liquids. hol, while in those having definite polymer moleIt should be mentioned a t this point that a third cules non-polar properties were to be expected. type of association through hydrogen, namely, Since then a great deal of additional evidence intramolecular association or chelation, may ochas been accumulated to emphasize this differ- cur, which has the result of producing a “normal” ence. For example, acetic acid shows no dipole behavior in a molecule which would otherwise moment in benzene solution whereas alcohols show abnormal behavior as a liquid and solvent. show an increasing polarization with increasing This type of association has been much studied by concentration. Alcohols also form glasses a t low Sidgwick6 and his collaborators. The conditions temperatures, which have been shown by Zacha- under which it occurs are well defined, and one of riasen2 to involve polymerization with hydrogen the important ones is apparently the possibility bond formation. It is not possible, of course, to of two configurations of the molecule with respect anticipate or predict in every case the type of poly- to the hydrogen, thus giving an opportunity for mer to be expected and that problem is one which resonance. still holds the greatest interest. The data from All of the types of association of interest in the which conclusions could be drawn have been avail- liquid state appear to involve hydrogen bonding. able for a long time; witness the excellent review The formation of polymers depends upon potenby Lassettrea based entirely upon the ordinary tial cross linkages or additional valences between data of physical chemistry. More recently a (4) G. F. Zellhoefer, M. J. Copley und C. S. Marvel, THIS JOURNAL, (1) W. M. Latimer and W. H.Rodebush, THIS JOURNAL, e,1432 60, 1337 (1938): G.F. Zellhwfer and M. J. Copley. ibid., Bo, 1343 (1920). (2) W.H.Zachariasen, J . Chcm. Phys., 8 , 168 (1936). ( 8 ) E.N. Louetuc, Chem. Rm..90, 269 (1987).
(1938). ( 5 ) N. V. Sidgwick, “The Electronic Theory of Valency,” Oxford Univudty Prua, 192P.
